Reproduce “MARCUS: An Event-Centric NLP Pipeline that generates Character Arcs from Narratives” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Sriharsh Bhyravajjula, Ujwal Narayan, Manish Shrivastava
MARCUS, an event-centric NLP pipeline developed by IIIT Hyderabad, quantifies and visualizes character arcs within narratives by tracking character interactions and their emotional circumstances. It demonstrated 71.5% accuracy in matching arcs to character pairs and 72.9% in identifying plot events with human readers, effectively operationalizing a theoretical literary concept.
Key insights
- Character transformation can be computationally represented as 'circumstance,' quantifiable by event-level sentiment and fine-grained emotions, offering tangibility to a theoretical literary concept.
- Distinguishing between a character's role as an 'actor' and an 'experiencer' provides a more nuanced understanding of their agency and how they are affected by narrative events.
- Event-centric analysis, leveraging semantic role labeling to identify specific participant roles, is crucial for understanding dynamic inter-character relationships and their impact on individual character journeys.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.